AI-supported evaluation application for product design assessment regarding automated production

In this article, software is developed to automatically evaluate the suitability of automated assembly of products. To achieve this, a product data structure is devised using the standardized AutomationML format, allowing for abstract product descriptions without the need for specific 3D CAD models. The developed evaluation system with its evaluation scheme is based on established methods. The basis of the textual description of the product and the abstraction of domain-specific knowledge with the help of fuzzy logic enables an innovative automated evaluation of the product design. By using fuzzy logic, subjective evaluation criteria can also be considered. For this purpose, a rule base comprising 21 evaluation criteria is established, resulting in 2430 unique rule combinations through effective grouping. In an initial evaluation of the developed software, three different fastening types from electrolyzer production were evaluated and compared in the context of instrument panels due to the manufacturing-like production. Here, the validation approach shows that the software can evaluate the suitability of the products for automation relative to the products under consideration. For a comprehensive validation, a comparison between the software evaluation and the evaluations of an expert panel is required, which will be carried out in later work.